MSFYOLO: Feature fusion-based detection for small objects

Abstract

At present, the effect of object detection algorithm in small object detection is very poor, mainly because the low-level network lacks semantic information and the characteristic information expressed by small object inspection data is very lack. In view of the above difficulties, this paper proposes a small object detection algorithm based on multi-scale feature fusion. By learning shallow features at the shallow level and deep features at the deep level, the proposed multi-scale feature learning scheme focuses on the fusion of concrete features and abstract features. It constructs object detector (MSFYOLO) based on multi-scale deep feature learning network and considers the relationship between a single object and local environment. Combining global information with local information, the feature pyramid is constructed by fusing different depth feature layers in the network. In addition, this paper also proposes a new feature extraction network (CourNet), through the way of feature visualization compared with the mainstream backbone network, the network can better express the small object feature information. The proposed algorithm is valuated on the MS COCO and achieved leading performance with 11.7% improvement in FPS, 17.0% improvement in AP, 81.0% improvement in ARS, and 23.3% reduction in computational FPLOs compared to YOLOv3. This study shows that the combination of global information and local information is helpful to detect the expression of small objects in different illumination. MSFYOLO uses CourNet as the backbone network, which has high efficiency and a good balance between accuracy and speed.